Sandberg1 recently argued persuasively that advancing technology puts some of the care currently delivered by anesthesiologists and/or certified registered nurse anesthetists (CRNAs) at risk for substitution by other providers. The hourly cost of monitored anesthesia care (MAC) delivered by an anesthesiologist and/or CRNA exceeds the cost prorated from the maximum societal cost per quality-adjusted life year.2 There is marked regional heterogeneity in the frequency at which anesthesiologists and/or CRNAs administer MAC.3,a For example, in Ontario, approximately 50% of colonoscopies at low-volume community hospitals are performed with MAC versus <5% at major academic hospitals.3 Propofol for MAC can be administered by closed-loop control.4–6 The SEDASYS system (Ethicon Endo-Surgery, Cincinnati, OH) integrates propofol delivery with patient monitoring, including automated assessment of response to auditory and tactile stimuli.7 Dexmedetomidine for MAC has a low incidence of respiratory depression and is relatively easy to titrate, with significant patient satisfaction.8 Providers other than anesthesiologists or CRNAs administer sedation for some procedures (e.g., colonoscopy and diagnostic radiology).9,10 At some facilities, registered nurses, who are taught and supervised by anesthesiologists, administer nitrous oxide, dexmedetomidine, or propofol for pediatric conscious sedation (e.g., endoscopy or diagnostic imaging).11,b
Among anesthesiologists and CRNAs surveyed in the United States, use of MAC averaged 27% and 29% of anesthetics and 16% and 24% of time, respectively.a These high percentages suggest that evolving technology and/or government regulation could substantively influence anesthesia workforce requirements12 nationwide.a However, these estimates may be biased because the survey was not a random sample of anesthetics and the data were incomplete. We used the 2006 National Survey of Ambulatory Surgery (NSAS) datac to obtain national rates for the percentage of ambulatory anesthesia operating room (OR) time that was sedation or MAC alone or after peripheral regional nerve block. Our analysis supplements the extensive report of these data published by the Centers for Disease Control.13,d
The study was performed using the publically available NSAS datae from the National Center for Health Statistics (NCHS). The survey population included all ambulatory surgery procedures performed in nonfederal hospitals and licensed freestanding ambulatory surgery centers in the United States in 2006. The survey examined patients' medical records, not facilities' billing data. To obtain unbiased national estimates, the NCHS assigned a weight to each surveyed case, based on (a) the reciprocal of the probability of sample selection, (b) nonresponse rate for similar cases, and (c) regional population. In addition, the NCHS stratified by facility type, specialty, and geographic region. We used the weights to calculate percentage estimates (Table 1). We used the weights and stratification to calculate standard errors. Data are reported as point estimates ± SE.
A case was considered an “anesthetic” if an anesthesiologist and/or CRNA was listed as present in the medical record. Anesthesiologists and CRNAs were not differentiated because survey respondents were not instructed what to enter for “anesthesia administered by” when a CRNA was medically directed by an anesthesiologist. For more detailed data, see References 13–15.
Type of anesthesia was coded into the following ranked sequence: (1) topical, intravenous, or MAC; (2) block, peribulbar, or retrobulbar (with or without option 1 also selected); (3) spinal or epidural (with or without 1 or 2 also selected); or (4) general (with or without 1, 2, or 3 also selected). There also were some cases in which “Other” or “None Specified” was checked. These represent all combinations of the options provided on the survey form. Our primary dependent variable was the percentage of OR time attributable to each type of anesthetic. We subsequently refer to 1 as “MAC” and 2 as “peripheral blocks.”
Analyses were repeated for specific common procedures to confirm validity of results. Another breakdown was by major therapeutic procedures.f
Analyses were also repeated by percentage anesthetics, rather than OR time. For brief procedures, American Society of Anesthesiologists' Relative Value Guide (ASA RVG) units are dependent mostly on basic units. Because MAC cases tend to be brief, the percentage of anesthetics that are MAC is an upper limit for the percentage of ASA RVG units that are MAC.
The Appendix shows the SUDAAN 10.0 code (Research Triangle Institute, Research Triangle Park, NC), with detailed explanations.
One-third (34% ± 2%) of ambulatory OR time nationwide with an anesthesiologist and/or CRNA was with MAC and/or peripheral block (Table 1). MAC cases alone comprised 29% ± 2% of OR time. Cataract surgery and colonoscopy under MAC and/or block accounted for 11% ± 1% of OR time. The remaining 23% was distributed over many procedures.
The percentages by case were 42% ± 3% for MAC and 47% ± 3% for peripheral block with/without MAC. Because percentages by cases were larger than by OR time (P < 0.0001), substantially more than one-third of ambulatory anesthesia workload (ASA RVG units) nationwide was MAC and/or peripheral block.
Our results supplement the already considerable analysis of these data published by the Centers for Disease Control.13,d Our work shows that MAC alone or after peripheral block accounts for a relatively high percentage of ambulatory anesthesia nationwide. Short term, anesthesia groups risk significant reductions in profit if some payers decide to reduce or curtail payments to anesthesiologists and/or CRNAs for providing MAC. Reasons for loss of payment for some procedures could be based on the cost2 and/or the observed heterogeneity3 in use. Technology4–8 facilitates substitution with less-expensive providers in situations in which the likelihood of conversion to general anesthesia, airway intervention, or cardiopulmonary management is low (e.g., see introductory text9–11,b). Accelerating the safe introduction of sedation technology and related information systems would facilitate workforce planning and anesthesiologists taking educational, research, and supervisory systems-based16 roles.1 A limitation of the national data is that MAC includes a continuum of sedation, including what may be considered general anesthesia, and depth of anesthesia is not known from the surveyors' records. Another limitation is that the use of MAC varies substantially among facilities3 and regions.a We did not analyze the data by region, because companies developing medications and devices for MAC respond to the national market, and anesthesia groups recruit nationally. Although individual regions may resist national trends, eventually economic and medical forces influencing national trends likely will prevail.
Individual anesthesia groups should plan based on our results, because the economic relevance of the survey findings will vary among groups. First, the NSAS data excluded patients with scheduled hospital admission or urgent surgery, thereby likely overestimating MAC percentages. However, the data excluded offices and nonlicensed ambulatory facilities, thereby underestimating MAC percentages. Because the data are from 2006 and technology advances over time, the degree of underestimation will likely increase progressively. Second, depending on case scheduling and method of reimbursement, some groups will be able to take advantage of technological advances in sedation to lower costs without reduced reimbursement, whereas others will not be able to do so. Consider a surgeon's series of cataract surgery cases performed in an OR on one day. Because compensation for registered nurses is less than for CRNAs, a group without incremental reimbursement would achieve net cost savings by substituting a registered nurse for a CRNA, except for uncommon medically challenging cases. In contrast, consider an OR with hand surgery cases, some with general anesthesia and others with sedation after nerve block. Hypothetically, with 2 such ORs, 1 OR could be sequenced with the general anesthesia cases done first and the regional plus MAC cases after. The other OR could have the opposite sequence. A CRNA and registered nurse could then swap ORs midday. However, statistical methods for such sequencing require no overlap between cases based on scheduled duration.17–20 The value of the statistical methods is to confirm a high probability of no overlap despite uncertainty in actual OR times.17–20 Without scheduled delays in both ORs as time buffers, the surgeons in the 2 ORs would often be delayed. Unexpected but reasonable requests by patients on the day of surgery for one type of anesthetic would cause even more delays and increase costs. Patients with obstructive sleep apnea and other medical conditions affecting MAC practice would also increase costs.
a Daugherty L, Fonseca R, Kumar KB, Michaud PC. An Analysis of the Labor Markets for Anesthesiology. RAND Technical Report, 2010:30–4, 71. Available at: http://www.rand.org/pubs/technical_reports/TR688/. Accessed February 17, 2011.
b The nurses receive didactic training, experience in operating room anesthesia care, start practicing with 1:1 direction by anesthesiologists, and generally are supervised 1:3. The Department of Anesthesia at the University of Iowa requested and the Iowa Board of Nursing approved that registered nurses administer propofol under the supervision of a physician. Approximately 90% of the 3600 sedations have been for children (<18 years). Available at: http://www.anesth.uiowa.edu/portal/portals/19/newsletters/2010%20Fall%20AnesthNews.pdf. Pages 8–10 accessed October 25, 2010.
c Survey of Ambulatory Surgery 2006, Revised Public Use Data File Documentation, May 2009:10–4. Available at: ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/Dataset_Documentation/NSAS. Accessed February 17, 2011.
d Reference 13. Available at: www.cdc.gov/nchs/data/nhsr/nhsr011.pdf. Accessed February 17, 2011.
e National Survey of Ambulatory Surgery Questionnaires, Datasets, and Related Documentation. Available at: http://www.cdc.gov/nchs/nsas/nsas_questionnaires.htm. Accessed February 17, 2011.
f Agency for Healthcare Research and Quality (AHRQ) Healthcare Cost and Utilization Project (HCUP) Procedure Classes define a major therapeutic procedure as a therapeutic procedure associated with operating room charges. Available at: http://www.hcup-us.ahrq.gov/toolssoftware/procedure/procedure.jsp. Accessed February 17, 2011.
g Research Triangle Institute. SUDAAN Language Manual, Release 10.0. Research Triangle Park, NC: Research Triangle Institute, 2008:80.
1. Sandberg WS. Barbarians at the gate. Anesth Analg 2009;109:695–9
2. Dexter F. Application of cost-utility and quality-adjusted life years analyses to monitored anesthesia care for sedation only. J Clin Anesth 1996;8:286–8
3. Alharbi O, Rabeneck L, Paszat LF, Wijeysundera DN, Sutradhar R, Yun L, Vinden CM, Tinmouth J. A population-based analysis of outpatient colonoscopy in adults assisted by an anesthesiologist. Anesthesiology 2009;111:734–40
4. Leslie K, Absalom A, Kenny GN. Closed loop control of sedation for colonoscopy using the Bispectral Index. Anaesthesia 2002;57:693–7
5. Fanti L, Agostoni M, Casati A, Guslandi M, Giollo P, Torri G, Testoni PA. Target-controlled propofol infusion during monitored anesthesia in patients undergoing ERCP. Gastrointest Endosc 2004;60:361–6
6. Stonell CA, Leslie K, Absalom AR. Effect-site targeted patient-controlled sedation with propofol: comparison with anaesthetist administration for colonoscopy. Anaesthesia 2006;61:240–7
7. Pambianco DJ, Vargo JJ, Pruitt RE, Hardi R, Martin JF. Computer-assisted personalized sedation for upper endoscopy and colonoscopy: a comparative, multicenter randomized study. Gastrointest Endosc 2011;7:765–72
8. Candiotti KA, Bergese SD, Bokesch PM, Feldman MA, Wisemandle W, Bekker AY. Monitored anesthesia care with dexmedetomidine: a prospective, randomized, double-blind, multicenter trial. Anesth Analg 2010;110:47–56
9. Vargo JJ, Cohen LB, Vargo JJ, Cohen LB, Rex DK, Kwo PY. Position statement: nonanesthesiologist administration of propofol for GI endoscopy. Am J Gastroenterol 2009;104:2886–92
10. Wachtel RE, Dexter F, Dow AJ. Growth rates in pediatric diagnostic imaging and sedation. Anesth Analg 2009;108:1616–21
11. Zier JL, Drake GJ, McCormick PC, Clinch KM, Cornfield DN. Case-series of nurse-administered nitrous oxide for urinary catheterization in children. Anesth Analg 2007;104:876–9
12. Schubert A, Eckhout G Jr, Tremper K. An updated view of the national anesthesia personnel shortfall. Anesth Analg 2003;96:207–14
13. Cullen KA, Hall MJ, Golosinskiy A. Ambulatory surgery in the United States, 2006. Natl Health Stat Rep 2009;11:1–25
14. Dexter F, Macario A. What is the relative frequency of uncommon ambulatory surgery procedures in the United States with an anesthesia provider? Anesth Analg 2000;90:1343–7
15. Rabbitts JA, Groenewald CB, Moriarty JP, Flick R. Epidemiology of ambulatory anesthesia for children in the United States: 2006 and 1996. Anesth Analg 2010;111:1011–5
16. Wachtel RE, Dexter F. Curriculum providing cognitive knowledge and problem-solving skills for anesthesia systems-based practice. ACGME J Grad Med Educ 2010;2:624–32
17. Dexter F, Traub RD. Sequencing cases in operating rooms: predicting whether one surgical case will last longer than another. Anesth Analg 2000;90:975–9
18. Dexter F, Epstein RD, Traub RD, Xiao Y. Making management decisions on the day of surgery based on operating room efficiency and patient waiting times. Anesthesiology 2004;101:1444–53
19. Dexter F, Ledolter J. Bayesian prediction bounds and comparisons of operating room times even for procedures with few or no historical data. Anesthesiology 2005;103:1259–67
20. Dexter F, Xiao Y, Dow AJ, Strader MM, Ho D, Wachtel RE. Coordination of appointments for anesthesia care outside of operating rooms using an enterprise-wide scheduling system. Anesth Analg 2007;105:1701–10
Using the National Survey of Ambulatory Surgery (NSAS) variables, a SASXPORT file was produced called “sas.xpt” with the fields listed in Appendix Table 1. The variable names are each 8 characters or fewer, matching SUDAAN requirements. The National Center for Health Statistics provides the variables to be used in SUDAAN codes.c
The SUDAAN code in Appendix Table 2 was modified slightly to create each of the rows and columns in Table 1 (e.g., limiting to cases of a specified CCS [Clinical Classifications Software]). The code in Appendix Table 2 calculates the percentage of total operating room (OR) minutes attributable to cases with anesthesia type of “topical/local,” “IV sedation,” or “MAC” (type = 1), among cases with an anesthesiologist and/or a certified registered nurse anesthetist (CRNA) present (AnesProv = 1). That is the result in the upper left hand corner of Table 1.
APPENDIX: SUDAAN CODE FOR READERS TO REPLICATE TABLE 1 AND/OR UNDERSTAND THE STANDARD ERRORS
The RATIO procedure is used. The location of the data file is defined with the DATA procedure. FILETYPE tells SUDAAN that the dataset is in SAS's XPORT format. DESIGN = WOR indicates that surgical cases were sampled without replacement (see Methods).c
Independent samples of subjects are drawn by strata, facility, and psudolog, respectively.c Strata include facility type, ambulatory surgery status of hospitals, facility specialty, and geographical region. The MISSUNIT option in the NEST statement effectively specifies that when only one surgical case is encountered within a facility, the variance contribution of that surgical case is estimated using the difference in that case's value and the overall mean value for all surgical cases for the strata.g
The TOTCNT statement has the same number of variables as in the NEST statement of Appendix Table 2C. The first item in this statement is the variable popfac, which specifies the number of cases at the facility from which a sample was taken for the survey. The second and third items in the TOTCNT procedure are SUDAAN defined options “_ZERO_” and “_MINUS1_.” “_ZERO_” instructs SUDAAN to generate a variable zero for every surgical case, making psudolog a stratification variable but with no variance contribution from any level of the variable psudolog. “_MINUS1_” instructs SUDAAN that sample units were selected with replacement for psudolog. This information is limited because “the design variables are not defined in any document. … To reveal their exact meaning would give information … we do not release to the public” (e-mail [personal] communication, November 12, 2010, Monica Wolford, National Center for Health Statistics).
The NSAS variable name for the weight of each case is called “weight.”c
The SUBPOPN statement specifies that the population of cases studied is those with an anesthesiologist and/or CRNA present.
The numerator of the ratio estimate (see below) sums the product of 5 variables over all NSAS cases: (case weight), (minutes of OR time), (0 for absence of OR time and 1 for presence), (0 for no anesthesia provider present and 1 for CRNA and/or anesthesiologist present), and (1 if the type of anesthetic is “topical/local,” “IV sedation,” or “MAC” and 0 otherwise). The denominator of the ratio sums the first 4 of those variables among all cases.
We created 4 examples (Appendix Tables 3.1–3.4). The SUDAAN code is identical to that of Appendix Table 2. The only change was that the data used were simulated, not NSAS data. Although in the NSAS dataset there are 73 strata, 72 strata were used in the simulations to have an even number of strata. Facilities were simulated within each stratum. Each facility was considered to have an even number of cases, half with the type of anesthesia being sedation and/or MAC and the other half being general anesthesia. The OR times were 60 minutes for all cases. Thus, the correct ratio estimate of total OR time with sedation and/or MAC was equal to 50%. For all simulations in Appendix Table 3, the SUDAAN point estimate was 50%.
Appendix Table 3.1 shows an example with an analytical answer for the SE. There are 72 strata. There is 1 facility within each of 72 strata. The weights are identical, equal to 100 cases for each of the 72 sampled cases. Thus, essentially, there are 36 sampled cases of 1s (for type of anesthesia is sedation and/or MAC) and 36 sampled cases of 0s (for type of anesthesia not sedation and not MAC). The sum of square deviations from the mean of 0.5 equals 18, where 18 = 36 × (1 − 0.5)2 + 36 × (0 − 0.5)2. The finite population estimate of the standard deviation equals 50%, where 50% = 100×
. Finally, dividing that 50% by the square root of 72 gives the finite population estimate of the standard error, which is 5.8926%. This is the value provided by SUDAAN, as listed in Appendix Table 3.1. The example highlights that the SUDAAN standard error estimate for the ratio procedure is based on the finite population.
Appendix Table 3.2 shows the impact of number of strata. With all other parameters constant, the standard error is proportional to 1/
. For example, when the number of strata is increased 4-fold from 18 to 72, the standard error is decreased 2-fold, which is the square root of 4.
Appendix Table 3.3 shows the impact of changes in each facility's population size (popfac). With all other parameters constant, the standard error is proportional to 1/
. For example, increasing popfac 10-fold from 100 to 1000 causes the standard error estimate to decrease 3.16-fold, where 3.16 =
Appendix Table 3.4 shows the impact of the weights. When the weights are the same for all cases, they cancel from the numerator and denominator, and so do not influence the estimate of the standard error. Consequently, Appendix Table 3.4 shows that the standard errors were the same whether each sampled case represented either 50 or 100 actual cases.
Franklin Dexter is the Statistical Editor and Section Editor for Economics, Education, and Policy for the Journal. This manuscript was handled by Steve Shafer, Editor-in-Chief, and Dr. Dexter was not involved in any way with the editorial process or decision.
Name: Emine O. Bayman, PhD.
Contribution: This author helped conduct the study, analyze the data, and write the manuscript.
Attestation: Emine O. Bayman has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Franklin Dexter, MD, PhD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Franklin Dexter has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
Name: John J. Laur, MD, MSc.
Contribution: This author helped conduct the study and write the manuscript.
Attestation: John J. Laur approved the final manuscript.
Name: Ruth E. Wachtel, PhD, MBA.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Ruth E. Wachtel approved the final manuscript.